The Influence of Institutional Regulatory Pressure on Nonprofit Hospital Audit Quality

Unlike their for-profit counterparts that focus on maximizing profit to the benefit of investors, nonprofit organizations have no investors, are mission-driven, and focus on benefiting the public. The absence of shareholders results in less stringent governance and accountability mechanisms for nonprofit organizations compared to investor-owned companies. Shortly after the enactment of the Sarbanes-Oxley Act (SOX), numerous scandals mirroring those in the public sector shifted stakeholder attention to nonprofit entities, with federal and state legislators and nonprofit-sector leaders calling for SOX-like nonprofit reform (Gilkeson 2007; Mead 2008). In response, Congress passed regulations including provisions affecting nonprofit entities and, for the first time in decades, the Internal Revenue Service (IRS) revised Form 990, Return of Organization Exempt from Income Tax, to include information designed to gauge accountability in nonprofit organizations and to specifically hold nonprofit hospitals accountable for their role in benefitting the community (U.S. Senate Finance Committee 2007).¹ Several states followed, passing or considering legislation addressing accounting, auditing, and/or disclosure requirements for nonprofit organizations. Pressure also arises from bond ratings agencies that now consider SOX-like governance and management practices when setting credit ratings for tax-exempt bond² financing (Moody's 2010).

The pressure to increase accountability, minimize potential liability exposure, maintain funding, and protect the organization's reputation have resulted in many nonprofit organizations revising their governance practices to voluntarily comply with SOX (Ostrower 2007; Venkataraman and Watkins 2008; Cohen 2012). Given voluntary compliance, opponents of increased nonprofit regulation argue that additional legislation requires justification for the allocation of scarce resources and careful rationalization and consideration of whether the desired nonprofit oversight and accountability have already been attained via improved audit quality (Keating, Fischer, Gordon, and Greenlee 2005; Saito and McIntosh 2010; Petrovits, Shakespeare, and Shih 2011).

In the nonprofit sector, lawsuits against audit firms on alleged audit failures are uncommon, and this decreased litigation risk can weaken the motivation of an audit firm to deliver a quality audit. Litigation risk provides an important incentive for audit firms to provide a quality audit with audit firms shifting their constrained resources to engagements of high litigation risk, potentially reducing diligence for clients with low litigation risk³ (Francis and Krishnan 2002; Li and Mande 2003; López and Peters 2010; Skinner and Srinivasan 2012). This is evidenced from the persistent and longstanding nonprofit audit quality issues of Circular A-133 audits under the Single Audit Act⁴ (López and Peters 2010). In the absence of litigation risk, auditor reputation becomes an important concern.⁵

The enactment of SOX and subsequent increased regulatory attention on the accountability of nonprofit organizations can shift the audit firms' judgment regarding the choice of previously acceptable audit procedures and increase the importance of reputational capital as an incentive for audit firms to improve audit quality to protect their reputation, maintain legitimacy, and ultimately survive in a highly regulated environment (Deis and Giroux 1992; López and Peters 2010; Francis 2011; Knechel et al. 2013). Research provides evidence of improved financial reporting quality of publicly traded companies after the enactment of SOX, supporting a fundamental shift in the audit firms' engagement practices (DeFond 2010; Lennox and Pittman 2010; Coates and Srinivasan 2014; DeFond and Zhang 2014). Although SOX primarily entails coercive rules for the audits of publicly traded companies, these rules also influence the expected normal audit practice in general and establish a cognitive base and normative pressure for the audits of nonprofit organizations (Carpenter and Feroz 2001), as future plaintiffs may argue that nonprofit directors and officers breach their fiduciary duties if they fail to act in accordance with the SOX reforms, which are perceived as “best practices” (Gere, Vitrano, and Schmelz 2005). The documented benefits in the for-profit sector may lead to a cascading effect that potentially increases the audit quality of nonprofit organizations through mimetic and/or normative pressures. Additional research is necessary for enhancing the understanding of institutional pressures on the audit quality of nonprofit entities (Tate 2007; Boone, Khurana, and Raman 2010; López and Peters 2010).

This study contributes to the extant literature by investigating whether regulatory and institutional pressures influence compliance with professionally endorsed accounting and auditing practices. Although prior research has examined changes in the audit quality of publicly traded companies after the enactment of SOX, limited evidence is available for understanding the effect of increased regulatory attention on the audit quality of nonprofit organizations. Such evidence is essential for advancing the ongoing debate on increasing nonprofit organizations' financial accountability. We provide some insight into whether significant changes to audit engagement practices and subsequent increased institutional regulatory pressures provide additional nonprofit oversight and accountability via improved audit quality. Specifically, we examine whether the auditor's propensity to report an internal control deficiency (ICD), a proxy for audit quality used in nonprofit research, has increased from the pre- to post-regulatory pressure (REGP) period for nonprofit hospitals. The uniqueness of nonprofit hospitals warrants a separate examination because this controls for industry effect, input and output market conditions, and regulatory pressures. Further, we extend the nonprofit audit quality literature by employing the absolute value of discretionary accruals (ABSDA), an observable proxy for audit quality commonly used in for-profit research, to examine whether accruals have decreased from the pre- to post-REGP period. To our knowledge, this is the first study to employ discretionary accruals as a proxy for audit quality in the nonprofit sector.

We find that the audit quality of nonprofit hospitals improved from the pre- to post-REGP period, suggesting that audit firms have responded to regulatory pressures and, as institutional theory purports, have enhanced their audit and engagement practices for the benefit of nonprofit hospitals and stakeholders. The results are consistent with an emerging body of research supporting an increase in audit quality of post-SOX nonprofit (López and Peters 2010; López et al. 2013) and for-profit (Lawrence, Minutti-Meza, and Zhang 2011; Coates and Srinivasan, 2014) organizations and provide evidence that SOX and the subsequent increased nonprofit regulatory attention have also enhanced the audit quality of nonprofit entities. Therefore, additional regulations may not be necessary for nonprofit entities because improved audit quality indicates that the desired nonprofit oversight and accountability may have already been attained.

The remainder of the paper is organized as follows. The theoretical framework and hypothesis are presented next, followed by an explanation of the research design and empirical findings. Finally, the limitation of this study, conclusions, and suggestions for future research are discussed.

Institutional theory⁶ considers regulatory oversight a form of coercive institutional pressure because government regulation requires conformity and accountability in exchange for the flow of resources into an organization (D'Aunno, Succi, and Alexander 2000; Krishnan and Yetman 2011). Organizations recognize the need for conforming with societal norms of acceptable practices and are inclined to succumb to institutional pressure to secure rewards, avoid punishment, and survive (Rollins and Bremser 1997). The basic tenet of institutional theory is that an entity strives to establish or enhance its legitimacy within a given environment by responding to internal and external pressures and adapting itself to emulate those around it (DiMaggio and Powell 1983; Meyer and Rowan 1977). Thus, an organization's structure, policies, and procedures respond to institutional pressures through conformity to institutionalized coercive rules and normative expectations (DiMaggio and Powell 1983; Meyer and Rowan 1977).

The passage of SOX resulted in the emergence of new regulatory structures for monitoring the audit work to secure audit firm legitimacy and ultimately subjugate auditing from a self-regulated professional activity to regulatory oversight under the Public Company Accounting Oversight Board (PCAOB) (Malsch and Gendron 2011). The PCAOB exposes audit firms to costly penalties, including financial penalties and revocation of licenses to practice, for poor-quality work (PCAOB 2004; DeFond 2010). An audit firm may face an impaired reputation and sanctions if a PCAOB inspection report reveals deficiencies in the audit work quality (PCAOB 2004; DeFond 2010). Hence, reputation and avoidance of costly penalties should encourage a rational audit firm to provide a quality audit to increase its legitimacy with external stakeholders (DeFond 2010; Lennox and Pittman 2010). Increased institutional regulatory pressure of nonprofit organizations after the enactment of SOX and the resulting changes to audit firm oversight are also potential impetus for improving the audit quality of all auditees (DeFond 2010).

Management implements internal controls to provide reasonable assurance on the achievement of organizational objectives in relation to effective and efficient operations, reliable financial reporting, fraud prevention, and compliance with applicable laws and regulations (OMB 2003). Audit quality is enhanced when an auditor objectively evaluates a client's control environment and resists client pressure against reporting an ICD. Unlike their publicly traded counterparts whose internal controls are under scrutiny after the enactment of SOX Sections 303 and 404, nonprofit organizations receiving federal funding have been subject to audits of internal controls over compliance and financial reporting since 1997. This situation provides a unique opportunity for examining a change in the reporting of ICDs between the pre- and post-REGP periods for nonprofit entities. Increased focus on internal controls and the potential penalties imposed by the PCAOB for poor-quality work not only alter auditors' judgment regarding the choice of previously acceptable audit procedures, but also place an increased emphasis on the reputational capital of the audit firm (López and Peters 2010). Therefore, audit firms can be incentivized to improve the quality of nonprofit audit engagements to protect their reputation, avoid financial losses, and ultimately survive in a highly regulated environment (Deis and Giroux 1992; Francis 2011).

Consistent with prior research (Keating et al. 2005; López and Peters 2010), we utilize an auditor's propensity to report an ICD (defined as the probability that an auditor will both detect and report a breach in the client's accounting system) as a proxy for audit quality (DeAngelo 1981). The reporting likelihood of ICDs among nonprofit organizations is expected to improve after the increased standard on internal control audits becomes the norm among the audit firms.

Improved audit quality due to institutional regulatory pressures may be reflected in the auditors' ability to constrain a client's earnings management, a commonly used proxy for audit quality of publicly traded companies. The extant literature on publicly traded companies concludes that low accruals indicate less aggressive earnings management and increased earnings and audit quality (Krishnan 2003; Francis 2004). While publicly traded companies manage earnings through manipulation of accounting accruals to increase firm value to meet or beat market expectations, nonprofit entities may manage earnings to a range around zero to minimize scrutiny by regulators, donors, and watchdog groups that may contend that nonprofit organizations with high earnings are not fulfilling their charitable mission (Leone and Van Horn 2005).

Empirical studies have uncovered incentives for nonprofit entities to manage earnings. Leone and Van Horne (2005) report the presence of nonprofit earnings management in a sample of nonprofit hospitals using three separate measures of discretionary accruals (i.e., third-party settlement liabilities, allowance for doubtful accounts, and the Jones model) and one measure of real activities management (i.e., expenditures on charity care). The authors find patterns of earnings management through significant use of discretionary accruals and a positive association between charity care expenditures and current income. Eldenburg, Gunny, Hee, and Soderstrom (2011) examine whether nonprofit hospitals manage income toward a benchmark of zero through real activity management. Their findings show that hospitals with strong motivation for managing earnings upward to achieve a positive income experience significant reduction in the expenditures of nonoperating and nonrevenue-generating activities. The hospitals also engage in significantly strong asset disposition management to avoid large positive net incomes.

Despite management's incentive for managing earnings, the role of the auditor is to limit the potential for client-opportunistic management of earnings to provide stakeholders with the assurance that the reported earnings reflect a reasonable and unbiased application of GAAP (Boone et al. 2010). Low accruals indicate decreased earnings management and increased audit quality. Significant changes to the audit engagement practices and subsequent increased institutional regulatory pressures in the post-REGP environment have provided additional nonprofit oversight and accountability suggestive of improved audit quality.

In sum, we argue that audit quality improves, in terms of a higher likelihood of reporting an ICD and a lower extent of accruals, with an increase in institutional pressure. Therefore, our hypothesis is as follows:

We use two measures of audit quality to test our hypothesis: ICD and ABSDA. Specifically, we examine the following model:

The dependent and independent variables are defined in Table 1 and discussed below.

Consistent with prior nonprofit audit quality research (Keating et al. 2005; López and Peters 2010), we utilize the auditor's propensity to report an ICD as a proxy for audit quality; that is, a quality audit suggests an increased probability of disclosure of an ICD (DeAngelo 1981; López and Peters 2010). ICD is an indicator variable equal to 1 for hospitals with a reported ICD, and 0 otherwise. Logistic regression analysis is used to test the dichotomous ICD variable.

ABSDA is also utilized as a proxy for audit quality, with lower accruals indicating less aggressive earnings management and higher earnings and audit quality (Francis 2004). Consistent with Cohen, Krishnamoorthy, and Wright (2008) and Doyle, Ge, and McVay (2007b), we use ABSDA because high-quality audit firms are more able to constrain earnings management. To estimate discretionary accruals, we utilize the residuals of the cross-sectional modified Jones model (Dechow, Sloan, and Sweeney 1995) as follows:

ACC denotes the total accruals calculated as the change in noncash current assets minus the change in current liabilities from year t−1 to year t minus depreciation expense for year t. ΔREV denotes the change in revenue from t−1 to t, ΔREC denotes the change in receivables from t−1 to t, PPE denotes gross property, plant and equipment, and the subscripts i and t denote firm and year, respectively. All variables are deflated by the beginning of year total assets. Discretionary accruals are estimated for each year of the model using an OLS regression.

PERIOD is an indicator variable equal to 1 for firm years 2006 to 2011, designated as the post-REGP period, and 0 for each firm year from 2001 to 2004, designated as the pre-REGP period.⁷ We start the post-REGP period with 2006 because state attorneys general and nonprofit-sector leaders have suggested nonprofit reforms that mimic SOX since 2005 (Mead 2008). As hypothesized, audit quality should improve from the pre- to the post-REGP era. Hence, we predict the sign of PERIOD to be positive when ICD is a proxy for audit quality and negative when ABSDA is a proxy for audit quality.

Consistent with prior research, we control for client and audit firm characteristics that may affect audit quality (e.g., Krishnan and Schauer 2000; Ge and McVay 2005; Keating et al. 2005; Doyle, Ge, and McVay 2007a; Lawrence et al. 2011; Petrovits et al. 2011). The control variables are audit firm type, size, complexity, risk, leverage, and audit firm change. Previous studies of nonprofit organizations (e.g., Petrovits et al. 2011) show higher ICD disclosures by regional audit firms compared to Big N firms. We control for the effect of audit firm type with AUDITFIRM, an indicator equal to 1 for non-Big 4 audit firms, and 0 otherwise, which predicts a positive association between AUDITFIRM and an ICD. While the for-profit sector commonly uses discretionary accruals as a measure of audit quality and finds that larger audit firms provide superior quality audits compared to smaller audit firms (DeAngelo, 1981; Francis, 2004; Lawrence et al, 2011), research employing discretionary accruals in the nonprofit setting is sparse. Hence, we make no prediction regarding the association between AUDITFIRM and ABSDA.

Prior research (Kinney and McDaniel 1989; Doyle et al. 2007a) suggests that hospital size (SIZE) has a negative effect on an auditor-reported ICD because large organizations tend to have strong internal controls leading to few auditor-reported ICDs and high earnings quality via low ABSDA. Therefore, we expect SIZE to be negative in the model. Complex transactions also increase the number and variety of operations leading to intricate accounting systems. As such, an organization with increased complexity (COMPLEXITY, measured by the revenue concentration index)⁸ are susceptible to internal control issues (Ge and McVay 2005), which increase the likelihood of an auditor-reported ICD and earnings management via increased ABSDA. Therefore, the coefficient on COMPLEXITY is predicted to be negative for an ICD and positive for ABSDA.

Auditee risk (López and Peters 2010; Petrovits et al. 2011) and financial leverage (DeFond 1992; Tate 2007) can affect audit quality. Circular A-133 audits require auditors to assess auditee risk. Auditees classified as low risk⁹ require less audit testing, which decreases the likelihood of an auditor-reported ICD. To control for this effect, we include an indicator variable, LOWRISK, which equals to 1 if the auditee is classified as low risk, and 0 otherwise. The coefficient on LOWRISK is expected to be negative when ICD is a proxy for audit quality. We note that the governance and accountability mechanisms for nonprofit organizations are less stringent than those for investor-owned companies. Therefore, auditors play an important role in monitoring the control environment of these organizations (Saito and McIntosh 2010). Lack of monitoring of low risk organizations may enhance opportunistic earnings management behavior. Thus, we expect LOWRISK to be positive when ABSDA is a proxy for audit quality.

Creditors require timely repayment of principal and interest regardless of a borrower's ownership structure. The existence of debt creates a creditor/borrower conflict that leads to increased monitoring demands (Jensen and Meckling 1976). We control for the effect of debt, LEVERAGE, measured via the ratio of long-term liabilities to total assets. Prior research suggests that changes in debt are related to changes in audit quality in publicly traded companies (DeFond 1992) and that nonprofit organizations with substantial debt are likely to hire quality audit firms to obtain an appropriate level of monitoring (Tate 2007). A nonprofit organization issuing tax-exempt debt may increase the litigation risk of an audit firm. However, a nonprofit entity may be subject to increased monitoring or compliance requirements from bondholders and external financing authorities. Hospitals with substantial debt are likely to have a quality audit; therefore, LEVERAGE is expected to be positively associated with an ICD and negatively associated with ABSDA.

To control for the effects of auditor turnover, we include an indicator variable, AUDITFIRMCHG, which is equal to 1 for audits involving a different auditor than in the prior fiscal year, and 0 otherwise. Given prior research findings suggesting that audit firm switching results in increased reporting of an ICD (Bedard, R. Hoitash, and U. Hoitash 2009; Feng and Elder 2017) and low accruals quality (Johnson, Khurana, and Reynolds 2002; J. Myers, L. Myers, and Omer 2003), we expect the coefficient of AUDITFIRMCHG to be positive for both measures of audit quality.

Table 2 provides information on the sample composition by year and REGP period. The sample is based on the Circular A-133 single audits obtained from the Federal Audit Clearinghouse (FAC) of the U.S. Census Bureau (https://harvester.census.gov/facweb/). The financial data are hand-collected from the hospitals' respective IRS Form 990, Return of Organization Exempt from Income Tax (https://www.guidestar.org). We exclude from our sample organizations with National Taxonomy of Exempt Entities codes other than E20 (hospitals and primary medical care facilities), E21 (community health systems), and E22 (general hospitals). We posit that increased institutional pressure after the enactment of SOX, the related PCAOB inspections, and increased scrutiny and enacted legislation directed at nonprofit organizations should motivate auditors to homogenize their improved audit and engagement practices for the benefit of nonprofit entities' audits. Thus, audit firms not registered with the PCAOB and those without at least one pre- and one post-REGP audit are eliminated. We also excluded from our sample audits without available financial data, integrated health systems not fitting the sample requirements, hospitals that are bankrupt or have their tax-exempt status revoked, and organizations not recognized as hospitals. Our final sample comprises 141 hospitals with 421 (37.5 percent) pre-REGP audit observations and 702 (62.5 percent) post-REGP audit observations. The number of annual observations remains relatively constant over the study period (Table 2).

Table 3 provides the descriptive statistics of the main variables for the full sample and for observations in the pre- and post-REGP periods.¹⁰ Comparison of the summary statistics between the pre- and post-REGP periods indicates that the mean of ICD is significantly larger in the post-REGP period (M_diff = 0.1163; p < 0.01). The mean of ABSDA is significantly smaller in the post-REGP period (M_diff = −0.0053; p < 0.01). These univariate results suggest improved audit quality from the pre- to the post-REGP period for both measures of audit quality, which provides preliminary supports to our main argument. As for the control variables, namely SIZE, LOWRISK, and AUDITFIRM, the univariate means are also significantly different between the pre- and post-REGP periods. We note that the number of non-Big 4 audits more than doubled from the pre- (12.4 percent) to post-REGP (26.2 percent) period. This is consistent with prior research findings (Feng and Elder 2017) on increased downward auditor switches in nonprofit organizations following the implementation of SOX. Most Big 4 audit firms might have eliminated their high-risk clients after the Enron Corporation scandal and/or passage of SOX, and the second-tier audit firms might have picked up these clients. In a similar fashion, the second-tier audit firms might have eliminated their high-risk clients and the smaller audit firms might have picked up these clients. This resulted in a substantial upgrading of client base, which filtered down through the firms based on size.¹¹ Our findings do not reveal a relation between audit firm change (AUDITFIRMCHG) and audit quality.

In our multivariate analysis, we control for differences in client and audit firm characteristics to examine any change in audit quality between the pre- and post-REGP periods and for Big 4 versus non-Big 4 audit firms. The highest pairwise correlations between the test variables and the following control variables are LEVERAGE (ABSDA: p = 0.200; ICD: p = 0.167), SIZE (ABSDA: p = −0.286; ICD: p = −0.071), and AUDITFIRM (ABSDA: p = −0.102; ICD: p = −0.170). Further, ICD and LOWRISK are significantly correlated (ICD: p = −0.110). We examined the variance inflation factors (VIF) and standardized error terms of the independent variables to assess multicollinearity in the estimated models. All the VIFs slightly exceeded 1 suggesting that multicollinearity does not affect the regression coefficient estimates and p-values.

Table 4 reports the logistic regression model results for ICD. Overall, the findings support our hypothesis on the positive association between PERIOD and ICD (β = 1.308; p < 0.001, Table 4, Column A), revealing an improved ICD reporting from the pre- to post-REGP period. The control variables of LOWRISK (β = −0.742; p < 0.001) and LEVERAGE (β = 1.079; p < 0.001) are significant and the signs are consistent with predictions. The findings of these control variables are consistent with prior research suggesting that low-risk audits are less likely to contain internal control exceptions (López and Peters 2010) and nonprofit organizations with substantial debt may be subject to increased monitoring and are likely to have a quality audit (Tate 2007). Table 4, Column B reports the estimates for the observations by post-REGP year to provide insight into the change in audit quality in the periods of increased institutional pressure. The findings show that the audit firms report increased ICDs beginning in 2006 (β = 0.929; p = 0.012) with continued significant improvement in subsequent years. This suggests that after the increased standard on internal control audits becomes the norm among the audit firms, the cascading effect may improve audit quality in nonprofit organizations. The peak increase of ICDs in 2007 and 2008 may be the result of an audit firm's provision of feedback that strengthens an auditee's internal control structure (López and Peters 2010). These findings are consistent with the descriptive statistics reported in Table 3, suggesting that the ICD audit quality measure improves from the pre- to post-REGP period.

As part of a robustness analysis, we examine the independent effects of hospital size, level of risk, and audit firm size on ICD. The results in Table 4, Column C provide further support for our hypothesis. The consistent positive and significant coefficients on PERIOD suggest that regardless of hospital size, level of risk, and audit firm size, audit quality improves from the pre- to post-REGP period.

Table 4, Column A shows that AUDITFIRM has a significant positive effect on ICD, demonstrating that non-Big 4 audit firms are more likely to report an ICD and, thus, provide higher quality audits than their Big 4 counterparts. This result supports the work of Keating et al. (2005) and Petrovits et al. (2011), who find that regional audit firms disclose the ICDs of nonprofit organizations at a disproportionately higher rate than their Big N counterparts. Figure 1 illustrates the findings. The mean ICD for both Big 4 and non-Big 4 audit firms in the pre- and post-REGP periods suggest that both audit firm types improve the quality of their audits. In the pre-REGP period, Big 4 and non-Big 4 audit firms report about 3.8 percent and 17.3 percent ICD, respectively. Big 4 and non-Big 4 audit firms report 14.1 percent and 25.5 percent ICD, respectively in the post-REGP period.

We present the OLS regression results for ABSDA in Table 5. The findings support our hypothesis. PERIOD has a significant and negative association with ABSDA (β = −0.004; p = 0.029, Table 5, Column A), indicating improved audit quality from the pre- to post-REGP period. The control variable, SIZE, (β = −0.015; p < 0.001) is significant and negative, suggesting that large organizations tend to have strong earnings quality, which leads to audit quality. LEVERAGE (β = 0.017; p < 0.001) also has a significant effect on ABSDA, but in an unexpected direction, indicating that hospitals with substantial debt have high ABSDA. One possible explanation for this finding is that hospitals with a large debt engage in earnings management to comply with related debt covenants. Table 5, Column B reports the estimates for the observations by post-REGP year to provide additional insight into the change in audit quality in the periods of increased institutional pressure. Unlike the findings for ICD that suggest a continuous improvement in audit quality, the results for ABSDA indicate improved audit quality in specific periods. The audit firms reported a significant decrease in ABSDA in 2008 (β = −0.007; p = 0.036) and 2010 (β = −0.008; p = 0.014). These two periods corresponded with institutional regulatory pressure where nonprofit hospitals faced increased scrutiny from the federal government, which led to a redesigned IRS Form 990 requiring increased reporting and disclosure requirements effective in the 2009 tax year.

Table 5, Column C presents the results of a robustness analysis examining the independent effects of hospital size, level of risk, and audit firm size on audit quality. The robustness test results show that the improvement of ABSDA is prevalent for hospitals that are small, have low audit risk, and that are audited by non-Big 4 audit firms.

The insignificant coefficient on AUDITFIRM in Table 5, Column A (β = 0.003; p = 0.150) suggests that Big 4 and non-Big 4 audit firms have similar audit quality (ABSDA), after controlling for client characteristics that may influence discretionary accruals. Figure 2 shows that the difference in ABSDA between Big 4 and non-Big 4 audit firms decreases further in the post-REGP period. In the pre-REGP period, the ABSDA of hospitals audited by Big 4 firms compared to those audited by non-Big 4 audit firms are 3.0 percent and 4.7 percent, respectively. In the post-REGP period, the ABSDA of hospitals audited by Big 4 firms relative to those audited by non-Big 4 audit firms are 2.7 percent and 3.2 percent, respectively.

Overall, our empirical findings show support for our hypothesis that improved audit quality is associated with increased regulatory pressure. Our results are also robust after controlling for various cross-sectional tests and sensitivity checks.

This study is not without limitations. While the discernable outcomes of an audit are the audited financial statements and the audit report, audit quality cannot be observed directly (Francis 2011; DeFond and Zhang 2014). Since explanations of audit quality revolve around either a simple dichotomy of the presence or absence of audit failure or gradations of audit quality along a continuum ranging from very low to very high, audit quality has different meanings for different stakeholders (Francis 2004, 2011). Numerous definitions of audit quality, each articulating the concept differently, have been examined. In our study, similar findings are obtained when audit quality is inferred from an auditor-reported ICD and the client's earnings management implicit in discretionary accruals. Thus, the results must be interpreted considering the proxies used for audit quality. As López and Peters (2010) note, use of internal control exceptions as a measure of audit quality may impact inferences because an audit firm may provide strong audit feedback that strengthens an auditee's internal control structure, thereby minimizing the existence of internal control exceptions over time. Conversely, large audit firms may reduce management's ability to manipulate earnings (measured by discretionary accruals), leading to strong audit quality (Lawrence et al. 2011). Future research can examine other aggregate and specific account (e.g., third-party settlement accounts, allowance for doubtful accounts, and charity care) discretionary accrual models and/or other proxies for audit quality in the nonprofit setting.

Although our regression models control for certain client characteristics that may influence audit quality, unaccounted for variation may still arise from differences in audit firm characteristics such as audit firm industry expertise, audit office size, audit firm tenure, audit fees, and differences in the clients' internal control systems, governance, or other financial and nonfinancial characteristics of the auditees. Additionally, our study is limited to nonprofit hospitals with Circular A-133 audits. Thus, our findings may not be generalizable to hospitals without Circular A-133 audits or to other nonprofit sectors. Future work can provide insight into this issue.

Despite these limitations, our results indicate a significant change in audit quality after the enactment of SOX and the subsequent increased institutional regulatory pressures imposed on nonprofit organizations.

In recent years, regulators and public interest groups have called for increased legislation for strengthening the accountability of nonprofit organizations and extending SOX-like dictates to these organizations. Others have argued that additional legislation requires careful rationalization of future mandates. Our study advances this ongoing debate by considering whether the significant changes to audit engagement practices and the subsequent increased institutional regulatory pressures have provided additional nonprofit oversight and accountability via improved audit quality. In this study, audit quality is inferred from examining internal control strength through an auditor-reported ICD and client-implied earnings management observed in discretionary accruals. To our knowledge, this is the first study to employ discretionary accruals as a proxy for audit quality in the nonprofit sector.

Our findings show that the audit quality of nonprofit hospitals (measured by ICD and ABSDA) improves from the pre- to post-REGP period, suggesting that audit firms have responded to regulatory pressures and, as institutional theory purports, enhanced their audit and engagement practices for the benefit of nonprofit hospitals and stakeholders. The results are in line with an emerging body of research supporting an increase in audit quality for post-SOX nonprofit (López and Peters 2010; López et al. 2013) and for-profit (Lawrence et al. 2011) entities and provide evidence that, given the increased audit quality, extending SOX-like regulation to nonprofit entities may not be critical at this time.

The choice of proxy for audit quality has an impact on the findings. When ICD acts as the proxy for audit quality, organizations with high risk and higher debt tend to have higher audit quality. The findings of these two control variables are consistent with prior research suggesting that risky audits are likely to contain increased internal control exceptions (López and Peters 2010), and nonprofit organizations with substantial debt may be subject to increased monitoring and are likely to have higher audit quality (Tate 2007). When discretionary accruals are a proxy for audit quality, the findings indicate that large organizations have earnings quality, suggesting audit quality. Contrary to expectations and the findings using ICD as a proxy for audit quality, when the proxy for audit quality is ABSDA, nonprofit hospitals with substantial debt have low earnings quality, indicating low audit quality. One possible explanation is that nonprofit hospitals with sizeable debt engage in earnings management to comply with related debt covenants.

Although both Big 4 and non-Big 4 audit firms improve their audit quality from the pre- to post-REGP period, the impact of audit firm size on audit quality depends on the choice of a proxy for audit quality. With respect to ICD as a proxy for audit quality, non-Big 4 audit firms provide higher pre- and post-REGP audit quality than their Big 4 counterparts. These findings are consistent with the pre-SOX nonprofit literature (Keating et al. 2005; Petrovits et al. 2011) and the results of López et al. (2013) suggesting that post-REGP nonprofit healthcare audits of the Big 4 audit firms are less likely to disclose internal control exceptions compared to the audits of smaller audit firms. However, the findings using ABSDA as a proxy for audit quality are consistent with previous for-profit research that generally concludes that large audit firms provide audit quality (Francis 2004; Lawrence et al. 2011).

Additionally, we examine whether the improved audit quality from the pre- to post-REGP period is the result of sample characteristic changes between the two periods. Our robustness analysis reveals that regardless of hospital size, level of risk, and audit firm size, audit quality improves from the pre- to post-REGP period when ICD serves as a proxy for audit quality. The findings using discretionary accruals as a proxy for audit quality show that the improved audit quality is prevalent for nonprofit hospitals that are small, have low audit risk, and engage non-Big 4 audit firms.

Future research can explore alternative proxies for audit quality and examine the linkage to fraud or earnings management in a nonprofit setting. Research opportunities also exist for investigation of additional audit firm and client characteristics that may influence audit quality. Our findings suggest that increased institutional regulatory pressure may have resulted in audit firms homogenizing their audit engagement practices to improve nonprofit earnings quality (i.e., similar audit quality is observed regardless of audit firm size). Future research can promote understanding of whether the extant literature's findings of the audit quality provided by large audit firms hold in this new regulatory environment.